Timeline for If a piece (classical or otherwise) has multiple sections with different keys, is "this piece is in key X" a valid statement? Why?
Current License: CC BY-SA 4.0
12 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 22, 2021 at 19:57 | comment | added | Morgan Kendal | @user45266 That is something being done indeed! Usually, a piece is divided into segments (which can vary in length) and these are individually investigated based on their correlation with every possible key. Trouble is, that is never the end goal with these systems, but rather a means to create a singular key label. That end goal seems very unproductive to say the least! | |
| Mar 22, 2021 at 18:10 | comment | added | user45266 | @MorganKendal Oooooh, working on key detection software, huh? That kind of thing can be very tricky, since even we humans sometimes disagree on the "correct" key! Not sure I know of any good method to tackle the issue of multiple keys being detected... would it be possible to have different keys turn up similarity scores based on how well they fit the music (C major is a 96% fit for this section of the music, and F major is a 45%fit, G minor is a 4.9% fit, et cetera...)? Just an idea off the top of my head, I don't know if it's feasible. Best of luck on your endeavor! | |
| Mar 22, 2021 at 14:33 | comment | added | bendl | @MorganKendal if you're doing any sort of "music automation" I'd recommend not using "songs" as the smallest "collection of sound". Songs can be far too complicated. I'd use sections of music that combined into larger ones. It makes sense for most music to talk about the key signature of one section of music, even in extreme cases like "Giant Steps". I'd say its very rare not to be able to break music up into discrete sections with describable characteristics, but common not to be able to apply those characteristics to the larger work. | |
| Mar 22, 2021 at 12:47 | comment | added | j-g-faustus | @MorganKendal Unambigious automatic key detection: Reinterpret the key as "the key signature that minimizes the number of accidentals". Not quite the same as the traditional meaning, but it has an objective metric. Less so if the music has many modulations, since you can reduce the number of accidentals by switching keys more frequently. But that's akin to compression, there is probably a way to describe an optimal number of key signature changes. It needs a 'cost' of key changes so it doesn't produce a key change for every single accidental. | |
| Mar 22, 2021 at 11:22 | vote | accept | Morgan Kendal | ||
| Mar 22, 2021 at 11:17 | comment | added | Morgan Kendal | I originally asked this question from a "music automation" standpoint - for automatic key detection, the standard procedure is to come up with a single key, even if the piece contains many modulations, without a clear single key label. I was mainly questioning that standard, and it seems to me that it is indeed a problem yet to be solved... | |
| Mar 22, 2021 at 8:34 | comment | added | Tim | And strangely, we tend to consider the last chord as representing the key, rather than the first. | |
| Mar 22, 2021 at 8:27 | comment | added | Tim | Nice list. I'd add 'Unforgettable', seems firmly in G to start, ends up in C. Written with a C key sig., though. And I can't figure out where it modulates! 4'33" - just checked, and it's written in 7#, modulating to 6b, but there's a version in key C for poor sight-readers. | |
| Mar 22, 2021 at 4:33 | history | edited | user45266 | CC BY-SA 4.0 |
deleted 7 characters in body
|
| Mar 22, 2021 at 0:31 | history | edited | user45266 | CC BY-SA 4.0 |
added 7 characters in body
|
| Mar 22, 2021 at 0:25 | history | edited | user45266 | CC BY-SA 4.0 |
added 4 characters in body
|
| Mar 22, 2021 at 0:20 | history | answered | user45266 | CC BY-SA 4.0 |